[rfc-i] <tt> vs HTML5

[cabo at tzi.org (Carsten Bormann)]

> We could easily fix this by getting rid of <tt> in the XML and replace
> it with <kbd>. Given that <tt> isn't in the v2 grammar, this change
> seems easy.

In an RFC, `kbd` is almost never what you want (we rarely standardize
what somebody has to type).  `code` almost always is.

We are migrating this from the v2 grammar <spanx style="verb"/>.  The
other styles -- except the undocumented "vbare" -- are very rare in RFCs
today, as they have to be edited out to avoid XML2RFC inserting
decorations around words in the TXT output that many authors find ugly,
while authors seem to rely on the double quote decorations for "verb".
I find that "verb" is used for ABNF fragments and other code samples*),
DNS names, tokens referring to text-based protocols such as JSON map
keys, some in-line formulae**), but also for names of RFC822-style
header fields as used to identify them in running text.  It also has
been used, mistakenly I'd say, to mark entire long-form citations.

> For historical reference, the design team discussed this back in October
> 2014 but didn't come to consensus and then kinda forgot about it. At the
> time, we were leaning against any presentational elements like <strong>
> and <em>, 

The presentational ones are `b` and `i`; `strong` and `em` are the
semantic ones.

Gr??e, Carsten

*) One authoring tool routinely inserts <spanx style="verb"
xml:space="preserve"> around those...

**) Lack of math support is probably the biggest remaining gap in
XML2RFCv3.  Not easy to fix...